The present invention relates to a network system, an OAM (Operations, Administration, Maintenance) regular test management server and a management method, and in particularly to an execution for scheduling a regular test, a test and a management of a testing result for the same configuration mentioned above.
Recently, a wide-area Ethernet service has been in widespread use, as a service for connecting inside sites distributed in a wide area by a network. Ethernet was originally a technique used for LAN (Local Area Network) and has been studied applying to a wide area network due to inexpensive cost for devices or any other reasons. One of problems of applying such a service to the wide area network was reliability thereof. For this reason of securing high reliability, an Ethernet OAM technique has been developed and standardized. The Ethernet OAM technique has been begun to widely apply to a network in business enterprises where the reliability is essential as well as to a wide area network for telecommunication firms.
The Ethernet OAM relates to a technique of streaming a test frame into a communication path through which user data also streams, to execute a continuity test and a performance test. In the wide-area Ethernet, a VLAN (Virtual LAN) is allocated to every user, and the test frame is streamed into the VLAN to execute the test on the communication path used also for the user data, so that a failure and performance monitoring can be executed accurately.
One of typically tested sections in Ethernet OAM is an access line for connecting between a core-network and a user inside site. The test is executed from an OAM device located in the core-network side to a device in the inside site when testing the access line. Several tens of thousands of devices are present in the inside site since they are corresponded to the number of access lines. In contrast, the number of OAM device is several in the core-network, therefore, it is required that the core-network side OAM device manages a number of devices in the inside site.
For a purpose of executing an OAM test, the OAM device is accessed by Telnet, SSH (Secure Shell) or the like, and a command is entered through a CLI (Command Line Interface). Normally, the number of session for the CLI to be able to connect simultaneously with a network device is as many as 3 to 16, and a number of the session is then limited to a number of the execution for testing the device in a large number of inside sites. In the case where the simultaneous connection of the session is realized by a number of CLI, a processing load becomes large, and it can be considered to be affected on a main signal. Therefore, it is hard that the number of session increases more than the present.
Presently, each of almost all the network devices incorporates the CLI, therefore, it is essential that the CLI is used for executing the OAM so that a large number of types of devices are targeted for management. In also the case where the network device incorporates a control IF (Interface) other than the CLI, the number of session is also limited.
In an applicative utilization method of the Ethernet OAM, there is a regular execution of the performance test, such as an LB (LoopBack), a DM (Delay Measurement) or the like. Unlike a CC (Continuity Check), the network device itself does not execute regularly the performance test such as the LB, DM or the like, but the performance test is executed by a chance given from a user. To this end, the performance test for the regular execution is realized by combining with, for example, a monitoring system. By executing the performance test regularly, it enables to monitor a segmentation of a monitoring section, an accurate performance measurement, etc. in high resolution. Therefore, there is an advantage that a failure handling can be speed up and a refund for an SLA (Service Level Agreement) violation can also be reduced.
When a monitoring result is submitted to a user as grounds of the SLA, it is important that the performance test is regularly executed for an accurate interval since information for a constant interval is required.
JP-A-2004-134969 discloses a system in which the OAM is executed regularly to suppress a size increase of a buffer memory used for receiving a control packet. But, the number of session of the CLI is limited for executing the performance test by using the above-mentioned Ethernet OAM. In consequence, it is difficult to execute a large number of tests regularly.
There has also proposed a method in which the test is executed for a shorter interval than a time in anticipation of a failure, as a method of executing the large number of tests within a limited execution session. But, a test time becomes longer in the failure and an execution interval varies, therefore, the regular execution cannot be carried through in an accurate interval.
JP-A-5-250186 discloses a system related to a task generation scheduling to generate a task for forcibly terminating when an execution time of the task is exceeded over a predetermined time. It is important to measure how long a delay is taken on the performance test, and the delay cannot be measured by causing the forcible termination. That is, it is important that the performance test should be executed through to completion even in the case where a processing time for the performance test becomes longer caused by the failure etc.